[ARROW-12529] [R] Writing to Parquet from tibble Consumes Large Amount of Memory

• Type: Bug
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Affects Version/s: 3.0.0
• Component/s: R
• External issue URL: https://github.com/apache/arrow/issues/28293
• Assignee: Unassigned
• Reporter: Jared Lander
• Created: 24/Apr/21 22:00
• Updated: 11/Jan/23 08:27

Description

When writing a large `tibble` to a parquet file, a large amount of memory is consumed. I first discovered this when using `targets::tar_read(obj)` to load in an object that had been saved in the parquet format. That particular object was an `sf` object with about 20 million rows and 26 columns. For a 5-6 GB object, memory ballooned by 22 GB.

I wrote the following code to test this using a regular `tibble`, not `sf`. In this test memory increases dramatically when writing, but not when reading, which I'm still trying to figure out.

library(arrow)
library(dplyr)
library(lobstr)
library(tictoc)n <- 10000000system('free -m')
tic()
fake <- tibble(
    ID=seq(n),
    x=runif(n=n, min=-170, max=170),
    y=runif(n=n, min=-60, max=70),
    text1=sample(x=state.name, size=n, replace=TRUE),
    text2=sample(x=state.name, size=n, replace=TRUE),
    text3=sample(x=state.division, size=n, replace=TRUE),
    text4=sample(x=state.region, size=n, replace=TRUE),
    text5=sample(x=state.abb, size=n, replace=TRUE),
    num1=sample(x=state.center$x, size=n, replace=TRUE),
    num2=sample(x=state.center$y, size=n, replace=TRUE),
    num3=sample(x=state.area, size=n, replace=TRUE),
    Rand1=rnorm(n=n),
    Rand2=rnorm(n=n, mean=100, sd=3),
    Rand3=rbinom(n=n, size=10, prob=0.4)
)
toc()
system('free -m')obj_size(fake)/1024/1024/1024system('free -m')
tic()
write_parquet(fake, 'data/write_fake.parquet')
toc()
system('free -m')system('free -m')
gc()
system('free -m')system('free -m')
tic()
fake_parquet <- read_parquet('data/write_test.parquet')
toc()
system('free -m')
obj_size(spat_parquet)/1024/1024/1024